|
Hobart
The AAD has a long term Adelie Penguin monitoring program located at Mawson
Station, Antarctica. Part of that program, involves monitoring selected nests
for presence of adults, chicks and eggs. Historically, this task has required
a biologist to visit each site as often as possible and visually count and record
the numbers of adults, chicks and eggs. Unfortunately, Antarctic weather and
sea ice conditions makes this very difficult if not impossible, so the AAD has
developed automated “Penguin Nest Cameras” which are programmed
to automatically take digital photographs of selected Adelie Penguin colony
nests, from the months of October to February each year, regardless of the weather.
Each season, each camera generates between 400 and 600 high images (about 2
gigabytes of data). There are currently 9 cameras at Mawson station, with potential
for a lot more at other locations around Antarctica. After each summer season,
the images are collected and a biologist sits down and has the task of looking
at every image to perform the counts. This process currently uses a collection
of programs
1) Adobe Photoshop : To create a template which marks the nest sites being studied,
and to overlay this template over all the images from each site, one at a time.
2) Microsoft Excel : To store the metadata for each nest.
3) Windows filesystem : To hold all the images and excel files in a sensible
structure based on the site and the season.
The process of using the different software programs to interpret the images in a consistent manner has not been strictly defined, and is open to variability based on how the tools are used, and the methods used by the biologist.
With increasing number of cameras, the task of classifying the many thousands
of nest images each season will grow and become very time consuming.
Thus, there is a clear need for an application that :
1) Provides a method of creating a template that defines the study nests.
2) Provides a clear and consistent way of accessing images for each site, season
and date/time then applying metadata to each study nest for that photograph
– for example the user could right click on a nest marker and a dropdown
properties menu would appear where they select and set the metadata. The metadata
would be stored in an XML file.
3) Provide a method for importing images from the camera memory card, into a
standard filesystem structure – that renames the files based on their
site and date/time information (from the embedded EXIF data) and stores them
in a consistent manner.
4) Works in a way that allows multiple people to work in parallel on different
image sets, and to cross compare & exchange data.
An integrated application, will allow the ability for anyone to re-visit any
site, season and date/time photo to see the nest metadata (adult count, eggs,
chicks, etc).
The nest camera is likely to be adopted by other research agencies. By having
a consistent data processing tool available to perform counts, and the metadata
collected and stored in a portable and exchangeable format, the different research
agencies can then easily cross compare and collaborate.
The application does not have to do image recognition, or have to visually interpret
any part of each image. That task is way too complicated – and is left
to the biologist. All that is required is an integrated method of accessing
the images, applying the data and storing it.
The AAD already has 2 seasons x 9 cameras worth of images and the associated
metadata, so there is plenty of test data to draw from.
Preferences (can be negotiated)
Source code Java J2SE 1.5 or similar
Development environment Java Builder 2005, Eclipse or other options
Code repository CVS
Data storage XML
AAD suggests the source code use the GNU General Public Licence so that it can be shared & developed with other research institutions worldwide.
Mr Kym Newbery
Phone Number 6232 3329
Email Address kym.newbery@aad.gov.au
Website www.aad.gov.au
Address 203 Channel Highway, Kingston
The estimated software difficulty rating is 4-4.5